Many households devote a large fraction of their budgets to "consumption commitments"-goods that involve transaction costs and are infrequently adjusted. This paper characterizes risk preferences in an expected utility model with commitments. We show that commitments affect risk preferences in two ways: (1) they amplify risk aversion with respect to moderate-stake shocks, and (2) they create a motive to take large-payoff gambles. The model thus helps resolve two basic puzzles in expected utility theory: the discrepancy between moderate-stake and large-stake risk aversion and lottery playing by insurance buyers. We discuss applications of the model such as the optimal design of social insurance and tax policies, added worker effects in labor supply, and portfolio choice. Using event studies of unemployment shocks, we document evidence consistent with the consumption adjustment patterns implied by the model. Copyright by the President and Fellows of Harvard College and the Massachusetts Institute of Technology.
